In order to understand the risk factors of legionella infection in children a case-control study with two kinds of controls was conducted. The stepwise regression analysis included age, sex, season and immunosuppressive therapy in regression equation. The non-conditional logistic regression analysis showed that age, passive smoking and touching soil were associated with legionella infection. The statistics analysis also showed that there were dose-response relationship between age, passive smoking and immunosuppressive therapy and the frequency of legionella infection in children. It was suggested that age, season, immunosuppressive therapy, passive smoking and touching soil were risk factors for legionella infection.